Positioning system for radiotherapy

A training system for training a dose based control module comprising:

• a first input for receiving a first image of a patient, the first image comprising tissue label data and dose specification data;
• a second input for receiving a second image of the patient, the second image including a designated part of the patient to be treated and being taken at a point in time later than the first image;
• a third input to receive a Ground Truth (GT) indication of a transformation vector specifying a correction of the position and/or orientation of the patient with which a dose delivered to the patient optimally approximates the dose as originally intended in accordance with the first image;
• a transformation vector generator unit to generate a plurality of transformation vectors;
• an image transformation unit to obtain respective transformed second images by applying the transformation vectors to the second image;
• an addition unit which for each transformed second image computes the corresponding GT-correction vector from the GT-indication of the second image and the transformation vector with which the transformed second image was obtained;
• a loss computation module to determine for each transformed second image a loss based on a difference between the corresponding GT-correction vector and the correction vector estimated by the dose based control module when the first image and the transformed second image are provided as input;
• an update module for updating parameters of the dose based control module based on the computed loss.

A method for training a neuromorphic dose based control module, comprising:

• providing a first image of a patient comprising tissue label data and dose specification data;
• providing a second image of the patient including a designated part of the patient to be treated, therewith also providing a Ground Truth (GT) indication of a transformation vector specifying a correction of the position and/or orientation of the patient with which a dose delivered to the patient optimally approximates the dose as originally intended in accordance with the first image;
• generating an augmented set of second images by applying mutually different geometrical transformation in three dimensional space to the second image and for each species of the augmented set of second images computing the Ground Truth for the required correction from the GT indication of the second image and the transformation applied to that species;
• with the dose based control module computing a respective output vector indicative for a correction of the position and/or orientation of the patient for each pair of the first image and a species of the augmented set of second images;
• for each species of the augmented set of second images computing a loss from the difference between the correction specified by the respective output vector and the respective GT;
• updating parameters of the dose based control module to reduce the loss.